00:34:35 · Molly O'Shea, Burkay Gür
[00:34:35] S0: You not going go home during weekends and you're still successful. [00:34:39] S2: Yes, imagine. Crazy. I think, like, there is a certain baseline, I think, that we try to be realistic about, like, what to expect. And so, that's part of the reversal, I think. Hey, like, can be a lot of, like, time crunch, you know, like, you know, we really push and we have to spend a lot of time working and sixty hours and eighty hours. And I think that, like, filters out maybe, like, certain, you know, certain people. So, I think looking for hunger is very important. And then I think, like, culturally, I would say anyone, like, who interacted with our team, like, on the customer side, they would probably attest to this is like our team is like pretty low ego, like pretty easy to work with. I think we look for like humility in general. Like, they gotta be really talented but at the same time, like not be a jerk about it. I think that's like pretty unique to find. And I think like anyone who's interacted with, like, our team probably knows that and that's, I think, why they really like us. Like, people are really easy to work with and, like, really friendly. I think you can have both at the same time. So, we try to filter for humility as well. [00:35:58] S0: What have been the biggest surprise challenges that you guys have come across? And not just the standard stuff, but it could just be something you had no idea you were gonna expect.
00:35:58 · Molly O'Shea, Burkay Gür
[00:35:58] S0: What have been the biggest surprise challenges that you guys have come across? And not just the standard stuff, but it could just be something you had no idea you were gonna expect. [00:36:09] S2: I'd say, like, being at the forefront, like, as a company is pretty hard. I think, one, you have to really want it. Right? Like, you have to want to build a big company. And two, you have to, like, keep up with it, right, which requires a lot of effort. So, I think in order to, you know, build a big startup, you really have to just be very committed and be very hungry. And I think, like, that's, you know, that definitely, that's a lot of work. You know, that takes a toll on you. I think Adam would say this, probably. You know? But, yeah, it just requires a lot of hard work. [00:36:53] S0: I don't know. I asked him what he's most excited about. He says he doesn't feel excitement. [00:36:59] S2: He's beyond that point. [00:37:00] S0: You're going to get there one day. No emotions.
00:37:00 · Molly O'Shea, Gorkem Yurtseven
[00:37:00] S0: You're going to get there one day. No emotions. [00:37:04] S1: Our industry is incredibly dynamic, right? When there's no good model for a month, everyone starts panicking, oh my god it's been one month, nothing new has happened. But the answer is just wait another month and something incredible is going to happen. It's been true since we started and that keeps excitement going, and that keeps people very encouraged that, okay, there's going to be the next model that's going to unlock these other use cases, which is going to unlock these other types of customers, and that has brought so much growth to our business, and so far it just keeps happening for us. [00:37:42] S0: Where have you been seeing some of the best models come from? Do they just come out sporadically? Is there any consistent theme between all of them?
00:37:49 · Gorkem Yurtseven
[00:37:49] S1: So, now we play a bigger role in the ecosystem because we are an infrastructure partner for the model providers, but we are also the API partner for the application layer or studios, people who are consuming the models. So we got a good sense of supply on demand on both sides. And we are in such a compute constraint moment right now that if you are coming up with a new model you really have to plan ahead of time, you have to make sure you have enough compute to run inference. If you are successful, you are likely not going to be able to procure it on your own, so it's usually a good idea to partner up with an inference provider like us to do your launch. So we work very intimate with the model providers. And for customers, they usually use multiple models at the same time, but they really care about model releases because they usually tie their product releases with model releases as well. So we keep in very close communication with them as well. And then we have to sometimes anticipate how much growth is gonna come on the model side to procure our own compute. To maybe do some investment on our side to enable the whole space. And someone really has to do that because otherwise all the compute is going to go to coding agents or whoever is doing By the bigger planning ahead and investing on compute a little bit we can enable both the model providers and the applications being built on it. So that has become our responsibility in the ecosystem these couple months because compute is so hard to come by.
00:39:29 · Burkay Gür, Molly O'Shea
[00:39:29] S2: I'll add a few more points. I think generative media is like very unique in the sense that in two ways. One, I think there's just more modalities. I think when people think of gen media, maybe they only think of video. Yeah. But like, you would have to think about text to image, image editing, video, three d, text to speech, you know, speech to text. I don't know if I said three d, like, gaming use cases, whatever. [00:39:56] S0: Three d, four d, five. [00:39:57] S2: Exactly. All of these. So, there's actually, like, many modalities within generative media. There's, like, I don't know, five to 10 companies in each of these sort of sub categories within Genmedia. So that's, like, one interesting aspect. So there's just, like, a lot more happening. There's more fragmentation. And then the other is going back to what I said about, like, diminishing returns. I think it's still, like, so early in the space that there's actually a lot more leapfrogging happening between more players because, like, we're not at, like, the 98 whatever percent. We're only at, like, I don't know, 50% of, like, you know, where we're ultimately going to go. So, there's actually more people leapfrogging. There's more diversity like, the number of labs. So, it's not just like, OpenAI Anthropic, OpenAI Anthropic, you know? It's more like there's like eight, you know, players and they're all like leapfrogging each other.
00:40:56 · Gorkem Yurtseven, Molly O'Shea, Burkay Gür
[00:40:56] S1: Is One big reason is you're not running out of data. There is like so much video data out there and so much new video data is being produced every day that no one claims to train all of that data, right? They have to pick the hard problem with training a video model is actually you have to filter out the interesting video data so that you make a good model. So if you need to put more data into the model, there is abundance and it's seemingly free. There's enough free video out there that it works very well. But also the problem space is bigger, right? You are trying to reason around like the whole world, physics, everything, so it's not as specific as coding or as specific as certain LLMs are trying to optimise around. So there's a lot more angles people can take to create a more interesting and appealing model. And it's been happening every other month almost. [00:41:55] S0: I think that's where the naming comes in too. [00:41:57] S1: There you go. You have to differentiate with your naming. [00:42:00] S0: I mean, they've got some funny names. [00:42:03] S1: Yeah. Happy Horse, Nano Banana. What do you like? There's Happy Oyster, the [00:42:09] S2: game model. [00:42:13] S1: There's MiniMax. Seedance and Seedream. Seedance, Seedream. I think those are... [00:42:19] S0: I think when Nano Banana had its moment and everyone was talking about Nano Banana for like three days, Even my dad was talking about Nano Banana. I was like, What the hell is going on here?